Adam Miles

memset of result buffer only clears first 25%

2

https://github.com/GPUOpen-Tools/gpu_performance_api/blob/30cd97819afd6f560a2dcd6847f5dde1fba08854/source/examples/dx12/dx12_color_cube/cube_sample.cc#L2311 I'm not sure why the sizeof(int) is there? Did the author think memset took a number of DWORDs to memset instead of bytes?

In two samples, the buffer had only one quarter of its size memset to zero due to erroneous divisions by sizeof(int). It probably didn't cause any problem at all, it...

The function is defined in Scene.h, but no such implementation exists anywhere in the repo.

Building against CUDA SDK

4

Hi, Was this code tested against the CUDA SDK, and if so, which version? I note that the C++ Includes setup includes the Intel OpenCL SDK, the CUDA SDK and...

Shader Model 6.6 Support

3

Could RGA be updated to include Agility SDK support so Shader Model 6.6 shaders built with an up-to-date version of DXC be supported for offline disassembly? Thanks, Adam

Compiling a compute shader for GFX1010 shows incorrect 'wave_size'

1

If you compile a compute shader with [numthreads(64,1,1)] the top of the ISA output says "wave_size(64)". However it's apparent from the generated asm that the code generated was designed for...

When using "--variable" as an argument to an inout function, 2 is subtracted not one

2

https://godbolt.org/z/73so7qzzW `void Func(inout uint byteOffset) { }` `uint PSMain(uint i : I) : SV_TARGET { Func(--i); // Subtracts 2... return i; }` The resulting DXIL has an "Add minus two"...